基于线性混合模型和NDVI阈值法的MODIS影像黄海浒苔监测

Monitoring of Enteromorpha prolifera in the Yellow Sea with MODIS image based on linear mixing model and NDVI threshold

水体富营养化等原因,导致浒苔(Enteromorpha prolifera)灾害自2007年在中国黄海海域频发,成为黄海最严重的生态灾害。卫星遥感具有大范围监测、瞬时优势,成为浒苔灾害最重要的监测手段之一。中分辨率成像光谱仪(moderate-resolution imaging spectroradiometer,MODIS)影像因其幅宽大、时间分辨率高、免费分发,成为浒苔业务化监测的主要数据源。由于空间分辨率(250 m)较低,混合像元的存在导致传统阈值法获取的浒苔面积误差较大。本文结合线性混合模型(linear mixing model,LMM)和归一化差值植被指数(normalized different vegetation index,NDVI)阈值法对250 m空间分辨率的MODIS影像进行浒苔面积提取。选择1个大区及其内部3个小区,以准同步5.8 m空间分辨率资源三号(ZY-3)卫星影像提取浒苔结果为准进行精度评价。发现NDVI阈值为0.04提取的浒苔像元对线性混合模型分解结果掩膜所得浒苔面积误差最小,大区及其3个小区的误差分别是7.86%、14.59%、-7.65%、-0.15%。应用本文提供方法可有效排除浒苔混合像元和非浒苔像元对浒苔面积信息提取的干扰,与阈值法相比大幅提高了反演精度,且在不同区域精度较稳定,可为浒苔生态灾害的处置决策和评估提供支撑。

Due to eutrophication, E. prolifera disasters have frequently occurred in China＇ s Yellow Sea since 2007, which have become the most serious ecological disaster in the Yellow Sea. Satellite remote sensing has the advantages of large-scale monitoring and instantaneous monitoring, and is one of the most important monitoring means of E. prolifera disaster. Moderateresolution imaging spectroradiometer （MODIS） image is the main data source of E. prolifera operational monitoring because of its large size, high temporal resolution and free distribution. There are many errors in E. prolifera area derived from NDVI threshold, because of mixed pixels in the coarse resolution （250 m） MODIS images. In order to solve this problem, we extracted the E. prolifera area from the MODIS image with the spatial resolution of 250 m by combining the linearspectral mixture decomposition method and NDVI threshold method. A large area and its three inner subareas were selected for the accuracy evaluation based on the data of E. prolifera extracted from quasi-synchronous ZY-3 satellite with 5.8 m spatial resolution. We found that the error of E. prolifera area, extracted from the linear mixed model with a NDVI threshold of 0.04, was the smallest, and the errors of the large area and its three subareas were 7.86%, 14.59%, -7.65% and -0.15% respectively. Hence, we provide a method that can effectively eliminate interference from the E. prolifera mixed pixel and non E. prolifera pixel, and greatly improve the inversion precision which is stable in different regions. This method can provide support for the management decision and evaluation of E. prolifera ecological disaster.